Table 5.
GLMM with binomial distribution predicting error rates by ordinal position
| Effect | Log-odds | SE | 95% CI | z-value | p | |
|---|---|---|---|---|---|---|
| LL | UL | |||||
| Model: Errors ~ ordinal position + (ordinal position || subject) + (ordinal position || category) | ||||||
| Fixed effects | ||||||
| Intercept | –2.25 | 0.16 | –2.57 | –1.93 | –13.89 | < .001 |
| Ordinal position | 0.05 | 0.05 | –0.04 | 0.15 | 1.06 | .291 |
| Percentage of erroneous trials | Ordinal position | |||||
| 1 | 2 | 3 | 4 | 5 | ||
| M | 10.42 | 12.08 | 11.39 | 11.53 | 13.19 | |
| SEM | 1.34 | 1.17 | 1.35 | 1.26 | 1.77 | |
Number of participants = 30; number of categories = 24; total N = 3600; SE = standard error; CI = confidence interval around the estimate; LL = lower limit; UL = upper limit; M = mean; SEM = standard error of the mean (Morey, 2008); erroneous trials = number of trials per ordinal position that were excluded due to errors (technical or answer-based). P-values are based on a Wald Z-test. Significant p-values of p < .05 are shown in bold